Distortion Analysis for Real-Time Data Collection of Correlated Fields in Randomly Distributed Sensor Networks

In this paper we consider the task of real-time reconstruction of data fields sensed by a network of smart sensors assuming that data is temporally and spatially correlated. We present an analysis framework to compute average distortion using realistic transmission schemes and network parameters. The existing work related to the real time data reconstruction problem has assumed regular grid deployments. In this paper we assume that sensor nodes are randomly deployed in the field thus making the problem significantly more realistic and challenging. We handle the randomness by dividing the field into n segments and randomly select k active nodes within each segment. This approach also facilitates the design of an efficient collision-free data transmission scheme to transmit data to the sink. We verify our analytical result by simulating data collection in fields that have different correlation coefficients. The analysis framework presented can be extended to different transmission schemes, node distributions and reconstruction methods.